neromous / RWKV-Ouroboros

This project is established for real-time training of the RWKV model.
50 stars 4 forks source link

之後有計畫新的訓練功能嗎? #5

Open win10ogod opened 1 year ago

win10ogod commented 1 year ago

之後有計畫新的訓練功能嗎? 例如: 1.類似AlphaGo Zero的自我对弈 2.多模態 3.huggingface數據集 等有潛能但項目沒有的功能

neromous commented 1 year ago

之後有計畫新的訓練功能嗎? 例如: 1.類似AlphaGo Zero的自我对弈 2.多模態 3.huggingface數據集 等有潛能但項目沒有的功能

只有多模态在考虑之列。

win10ogod commented 1 year ago

AlphaGo Zero

您認為AlphaGo Zero的思路能否用於對話式模型的學習([類似Google 的Gemini 模型),或RWKV模型用於設計芯片的潛能如何?

win10ogod commented 1 year ago

AlphaGo Zero

您認為AlphaGo Zero的思路能否用於對話式模型的學習([類似Google 的Gemini 模型),或RWKV模型用於設計芯片的潛能如何?

也許RWKV模型用在機械設計也會是不錯的方向呢

neromous commented 1 year ago

AlphaGo Zero

您認為AlphaGo Zero的思路能否用於對話式模型的學習([類似Google 的Gemini 模型),或RWKV模型用於設計芯片的潛能如何?

也許RWKV模型用在機械設計也會是不錯的方向呢

看路标, 目前规划太大的功能不现实。 当前能将简单业务闭环已经需要花很久的时间实现。